AI’s environmental paradox: Progress at a cost

Yet behind the innovation lies a sobering reality: AI consumes vast amounts of energy, requires huge volumes of water for cooling servers, and generates electronic waste that is often non-recyclable.

Consider the scale. Training a single large AI model can consume as much electricity as hundreds of households use in a year. Data centres, the backbone of AI, are expanding rapidly. In regions such as Ashburn, Virginia, now known as the Data Centre Alley, where data centres are concentrated, communities have felt the pinch in their water bills.

Cooling servers is water-intensive, on average, consuming about nine litres per kilowatt-hour. When you multiply that across thousands of servers running continuously, the environmental cost becomes staggering. According to a study by The Conversation, this heavy usage is also said to affect air and water quality, noise levels, land use and energy costs.

These figures raise urgent questions of ethics and accountability. From a cost-benefit perspective, do the gains of AI outweigh its environmental costs? And if they do, who bears responsibility for ensuring that the balance is fair?

Every prompt we type into Copilot, Gemini, or ChatGPT carries an unseen environmental footprint. The convenience of instant answers, predictive analytics, or creative content is not free; it is subsidised by energy grids, water systems, and communities that may never directly benefit from the technology.

As a student at the University of Lancashire in the UK, currently studying Business Analytics and AI, this debate feels especially close to home. Just this week, our lectures focused on the impact of data and AI, not only in terms of efficiency and innovation, but also in terms of sustainability and ethics. It is striking to see how the same algorithms that promise breakthroughs in business can simultaneously strain the very resources we depend on for survival.

We observed that about 1.1 billion people lack access to water, and about 2.4 billion, the majority in sub-Saharan Africa, suffer water and sanitation issues, causing exposure to cholera, typhoid and other water-borne diseases.

Yet, paradoxically, AI also offers powerful tools for sustainability.

Through satellite monitoring, AI can track deforestation in real time. With smart grids, it can optimise the energy use, reducing waste and integrating renewable sources more efficiently.

AI-driven early warning systems are already helping communities prepare for floods, droughts, and wildfires, saving lives and resources. In agriculture, AI models are guiding farmers to use water more efficiently and reduce pesticide use, directly contributing to environmental protection.

This duality, AI as both a strain and a saviour, is the paradox of our time. The challenge is not whether AI should exist, but how it should be governed.

Accountability frameworks, transparent reporting, and investment in greener infrastructure must become non-negotiable. Tech companies should be required to disclose the environmental footprint of their models, just as industries disclose carbon emissions.

Governments must incentivise the development of green AI in which efficiency and sustainability are built into design rather than treated as afterthoughts.

There are encouraging signs. Some firms are experimenting with renewable-powered data centres, while others are exploring liquid cooling systems that reduce water use. Researchers are working on smaller, more efficient AI models that deliver comparable results without the massive energy drain.

However, these efforts remain scattered and voluntary. Without clear regulation and global standards, progress will be uneven, and the environmental costs will continue to mount.

As we continue to reflect on World Environment Day, the question is not whether technology can save us, but whether we will demand that it does so responsibly. AI is here to stay. Its potential to drive sustainability is immense, but so is its capacity to undermine it if left unchecked.

Janet Sudi -Maina is a Communications & AI Expert